The dispensability of (merely) intentional objects

Abstract

The ontology of (merely) intentional objects is a can of worms. If we can avoid ontological commitment to such entities, we should. In this paper, I offer a strategy for accomplishing that. This is to reject the traditional act-object account of intentionality in favor of an adverbial account. According to adverbialism about intentionality, having a dragon thought is not a matter of bearing the thinking-about relation to dragons, but of engaging in the activity of thinking dragon-wise.

1 Ontological commitment

The topic of ontological commitment raises two distinct kinds of philosophical issue. The first concerns what we should be ontologically committed to, which putative entities, or ontoids, we should accept as genuine ones.¹ The second concerns what it is for us to be ontologically committed to something. In a way, the first issue asks for the extension, whereas the second for the intension, of “exists” or “is” (treated for the sake of exposition as a first-order predicate). We may say that the former is the subject matter of ontology, the latter that of meta-ontology.²

Meta-ontology, so understood, has been dominated for a long time by Quinean ideas about ontological commitment, according to which for us to be ontologically committed to something is for us to quantify over it. More recently, a truthmaker-based alternative has been gathering momentum, a central aspect of which is that predicate terms are just as ontologically committing as subject terms.

Ontology, as practiced in the past half-century or so, has been concerned with a number of logically (though seldom dialectically) independent issues regarding certain putative entities whose existence is relatively controversial. Least controversial are putative entities in the category of “concrete particulars” (or “individual objects”).³ To be sure, for any type of putative entity, there is a philosopher who denies its existence, and so even concrete particulars of every stripe have met with ontological resistance. But the more controversial putative entities typically belong to the categories of abstract universals and abstract particulars, such as properties, numbers, values, events, and states of affairs.

My concern in this paper will be with another ontological (as opposed to meta-ontological) issue, namely, of whether we should accept the existence of intentional objects, or more accurately (as we will see momentarily) merely intentional objects. The intentional object of a thought is that-which-one-thinks-about. Thus, when I think of Vienna, Vienna is the intentional object of my thought. More generally, the intentional object of an intentional act or state is that which the act or state is of or about. Without pretending that this is an analysis, we may say that x is an intentional object iff there is a y, such that y is an intentional act and x is that which y is of or about.⁴

The term “intentional object” suffers from a number of ambiguities we would do well to undo before going any further. First, some use the term to denote any objects which happen to be such that some intentional act is directed at them, while others use it to denote objects that would not exist unless an intentional act was directed at them. In the first usage, the existence of an intentional act directed at x is a necessary condition for x’s being an intentional object; on the second, it is a necessary condition for x’s being simpliciter. I will align my use of “intentional object” with the former usage and employ “merely intentional object” to correspond to the latter usage. The way I use the term, x is a merely intentional object of act y iff (i) x is an intentional object and (ii) the following counterfactual holds: if there was no y, then there would not be x. At the same time, some use “intentional object” to cover only objects that are actually targeted by some intentional act, while others use it to cover also objects that could potentially be targeted by some such act. In other words, although the term is sometimes used so that x is an intentional object iff there is a y, such that y is an intentional act and x is that which y is of or about, it is also sometimes used so that x is an intentional object iff possibly, there is a y, such that y is an intentional act and x is that which y is of or about, where the modal operator is construed either metaphysically or nomologically, depending on one’s conception of the modality involved in potentiality. Here I will use “intentional object” in a way that requires an actual act and “potential intentional object” in a way that requires only a possible one.

The question I want to address here, then, is whether we should commit to the existence of merely intentional objects. This is one specific issue within ontology. But it is also more than that. For all other putative entities, whatever else they might be, are in addition intentional objects, or at least potential intentional objects. Properties, numbers, and events are very different from each other, but they are all thinkables—they can all be contemplated or entertained. Trivially, if the philosopher cannot even contemplate or entertain an F, then Fs are not the kind of thing the philosopher could end up ontologically committing to. Thus being a potential intentional object is a necessary condition for being something we are ontologically committed to. Putative entities whose existence we ratify are the winners in the game of ontological commitment, those we reject are the losers; but to even enter the game of ontological commitment, a putative entity must be a potential intentional object.

There may be entities of which humans cannot in principle think—entities we are cognitively closed to. But arguably, for any putative entity, there is some possible cognizer by whom they are thinkable. There are no putative entities an omniscient god cannot think of, and such a god is arguably possible. If so, the category of potential intentional objects is the highest and most inclusive ontological category. It is the only ontological category which is not a sub-category of any other ontological category. Another way to put it is to say that “potential intentional object” is the only count noun which cannot function as a sortal.

What this means, in any case, is that the question of the existence of potential intentional objects is the most general that could be raised within ontology (still as opposed to meta-ontology). If there is an argument for committing to all potential intentional objects, including merely intentional objects, this would produce in one fell swoop an argument for the existence of properties, numbers, values, events, and any other controversial or uncontroversial putative entity ever debated by philosophers.⁵ It should be stressed, however, that the converse does not hold: an argument against ontological commitment to merely intentional objects would not be ipso facto an argument against commitment to properties, numbers, etc. At most, such an argument would show that if properties, numbers, etc. exist, it is not in virtue of being intentional objects that they do.

The paper divides into two broad parts. In Sects. 2–5, I defend an eliminativist view of merely intentional objects, based on a certain account of the nature of intentionality. In Sects. 6–9, I defend that account of intentionality, as well as its necessity for eliminating merely intentional objects, from a number of objections.⁶

2 Commitment to (merely) intentional objects

It is a fair comment, I think, that pre-philosophically the more natural view is that there are no merely intentional objects.⁷ There are no dragons and no fairies, hallucinated butterflies are unreal and Cinderella is “made up.” I want to claim, more generally, that there is no pressure from the phenomena to admit merely intentional objects. The only pressure comes from certain philosophical doctrines.

What are “the phenomena” here? One phenomenon is that Jimmy is seeing a tree right now. This phenomenon clearly does not require the postulation of a merely intentional object. The intentional object of Jimmy’s seeing is that which Jimmy sees. That which Jimmy sees is the tree. So, the intentional object of Jimmy’s seeing is the tree. The tree’s status as an intentional object depends on Jimmy’s seeing, but its existence does not. Therefore, the tree is not a merely intentional object. In fact, there is nothing ontologically special about this tree: it does not become fundamentally different in the wake of Jimmy’s seeing.

A more problematic “phenomenon” is that Johnny is right now hallucinating a tree, in fact a tree qualitatively indistinguishable from the tree Jimmy is seeing. What is the intentional object of Johnny’s hallucination? The natural thing to say here is that Johnny’s hallucination has no intentional object. Its intentional object would be the tree which Johnny is hallucinating, but since there is no tree, there is no intentional object.

Thus there is nothing about the phenomena that forces us to admit merely intentional objects. And yet philosophers have felt a strong pressure to posit such. We can appreciate the pressure by considering the following argument:

• (1) If x is a tree-seeing, then there is a y, such that y is the intentional object of x;

• (2) Tree-seeings and tree-hallucinatings belong (other things being equal) to the same intentional kind (or are tokens of the same intentional type);

• (3) Intentional acts/states that belong to the same intentional kind have the same kind of intentional object (same intentional act type, same intentional object type); therefore,

• (4) If x is a tree-hallucinating, then there is a y, such that y is the intentional object of x.

Once we have reached this conclusion, we enter the can of worms of offering an ontological assay of “hallucinata.” Since the hallucinated tree is ex hypothesi not an existing concrete, physical, spatio-temporal tree, it must be something else. Three main options have presented themselves to philosophers. The first is that the hallucinated tree is an abstract object (see Salmon 1988). The second is that it is a concrete object, but a mental one (see Jackson 1977). The third is that it is a non-mental concretum, but a non-existent one (see Parsons 1980; Priest 2005).

These three options—construing hallucinated trees as abstracta, as mental concreta, and as non-existent non-mental concreta—interact in different ways with the ontological issues mentioned in Sect. 1. As noted there, if one embraced merely intentional objects, one would have thereby embraced properties, numbers, values, events, and other problematic putative entities, since the latter are thinkable. Thus, a philosopher unimpressed by Platonist defenses of mathematical entities, but who agreed that merely intentional objects were mental concreta, would end up with the view that numbers are mental concreta; someone impressed by eliminativist arguments about events, but who thought that merely intentional objects were abstracta, would be saddled with the view that events were abstracta; a sympathizer of moral fictionalism who held that merely intentional objects must be construed as non-existent concreta, would be landed with the view that moral values were non-existent concreta; and so on and so forth.⁸ Some of the possible combinations here would be more natural than others. But it should be clear how the two issues interact.

I said that the assay of merely intentional objects is a can of worms. There are quite familiar and truly outstanding liabilities—ontological, epistemological, and phenomenological—associated with saying that merely intentional objects are abstracta, or mental concreta, or non-existent non-mental concreta.⁹ To avoid this can of worms, we would have to reject one of the premises of the above argument. It is hard to deny the first premise, but the second and third are philosophical doctrines that should be revisited in light of the can of worms to which they lead. The second premise is effectively rejected by disjunctivists.¹⁰ The third premise is characteristic, perhaps definitive, of the so-called act-object account of intentionality, according to which intentionality is a relation between an act and an object. It is this third premise I wish to reject.

3 Theories of intentionality

There are many ways to organize the logical space of theories of intentionality. In this section, I propose one useful given our purposes. The central distinction is between theories that cast intentionality as a relation between an act and an object (act-object theories) and theories that do not; each then divides into two main varieties.

According to act-object theories, being in an intentional state (or performing an intentional act) involves, constitutively, bearing a relation to an intentional object.¹¹ Thus, thinking about Vienna involves, constitutively, bearing an aboutness relation to Vienna. More generally, representing x involves, constitutively, bearing a representation relation to x.¹²

We may divide act-object theories into two main varieties. The first requires that the intentional object exist, the other does not. Call the former the “existent-object view” and the latter the “object view.”¹³ This difference comes out in acute cases of misrepresentation. On the existent-object view, thinking of dragons is a matter of bearing an of-ness relation to existing dragons, albeit abstract ones or concrete mental ones. (These are two versions of the existent-object view.) On the object view, thinking of dragons is a matter of bearing a relation to physical, flesh-and-blood dragons, albeit non-existent ones.

The alternative to the act-object theories are theories according to which being in an intentional act/state does not involve constitutively (though it may involve contingently) bearing a relation to an intentional object. On this view, thinking about something is never constituted by bearing a relation to that-which-one-thinks-about. I think of this view as a type of adverbialism about intentionality: to think about Vienna does not involve (constitutively) bearing an aboutness relation to Vienna, but rather engaging in the activity of thinking in a certain way—Vienna-wise. More generally, for any x, representing x does not involve constitutively bearing a representation relation to x; what it involves constitutively is representing x-wise.¹⁴

The obvious challenge for this view is to explain what it means to represent x-wise. Here is where varieties of adverbialism differ. To my mind, there are two main varieties (though in Sect. 9 we will encounter a third variety), which I call inferentialist adverbialism and phenomenological adverbialism. According to inferentialist adverbialism, thinking Vienna-wise is a matter of being in a mental state that has a “Viennese” inferential or functional role. There are a number of different accounts of inferential role (see Field 1977; Block 1986; Brandom 1994), and each could be used in a different version of this kind of adverbialism.¹⁵ According to phenomenological adverbialism, thinking Vienna-wise is a matter of being in a mental state that has a Viennese phenomenal character. Here it is important to distinguish two different notions of phenomenal character in the extant literature. One is the notion of phenomenal character as a kind of sensuous quality, the other that of phenomenal character as whatever property the explanatory gap concerns. The phenomenological adverbialist most certainly ought to employ the second notion, as it is unclear that thoughts about Vienna have a proprietary and universal phenomenal character in the first sense.¹⁶

Figure 1 illustrates the taxonomy of theories of intentionality that I have offered.¹⁷ It should be stressed that the above theories are supposed to apply not to all intentionality, but only to so-called underived or original intentionality. It is commonly thought that only some intentionality is original or autonomous, whereas other intentionality is derivative or inherited and occurs only by courtesy of the first type of intentionality. It has typically been claimed that linguistic intentionality derives from mental intentionality, say via the intentionality of speakers’ communicative intentions (Grice 1957; Schiffer 1982). More recently, some have argued that only conscious or phenomenal intentionality is original and underived (Kriegel 2003; Strawson 2004).¹⁸ All above versions of the act-object theory and the adverbial theory should be understood to apply to original or underived intentionality. The status of derived intentionality is not central to the issues that will concern us (for reasons we will encounter momentarily).

Fig. 1

Theories of intentionality

4 Adverbialism and merely intentional objects

We saw that the argument presented in Sect. 2 embroiled the philosopher in a can of worms, the assay of merely intentional objects. But Premise 3 in the argument is a commitment of act-object theories of intentionality only. Adverbialist theories deny the premise, and can therefore adopt the simpler and more natural view that some intentional acts/states do not have an intentional object. For they hold that hallucinating a tree is just a matter of hallucinating tree-wise and does not involve bearing a relation to a hallucinatum. Likewise, thinking about dragons is just a matter of thinking dragon-wise and does not involve bearing any relation to dragons. More generally, representing x is a matter of representing x-wise rather than of bearing a representation relation to x. In such a view, there is no ontological commitment to what Quine called entia non grata as part of accounting for the existence of objectless intentional acts. The acts exist, the objects may or may not.

We can appreciate this by considering that quantification over, or any reference to, merely intentional objects can be dispensed with through a combination of pleonastic transformations and paraphrases. Thus, “the dragon I dreamt of last night was yellow” would be pleonastically transformed into “I dreamt of a yellow dragon last night,” and the transform would be paraphrased into “I dreamt yellow-dragon-wise last night.” While the original sentence quantifies over dragons, and the midway transform arguably calls for dragons as constituents of truthmakers, the final paraphrase does neither.

It is possible to avoid commitment to merely intentional objects, then, but the price, if such it is, is the adoption of adverbialism about intentionality. It strikes me that recent discussions of intentionality have treated this web of issues rather glibly. The attitude is that intentionality can be construed as a relation between an act and an object but—not to worry—sometimes the object does not exist, where this is understood to involve no ontological commitment rather than to involve commitment to something like non-existent objects. “Intentional relations are special like that.” The only way this could make sense, however, is if a two-place relation could be instantiated even when there are not two relata. Yet nobody thinks it remotely plausible that a monadic property could be instantiated even when there is no entity that instantiates it, e.g., that squareness could be instantiated even if there are no squares. The same sense of absurdity should attach, I contend, to the parallel claim about relations: just as a monadic property cannot be instantiated in the absence of an instatiator, so a relation cannot be instantiated in the absence of relata.¹⁹

There is here a fairly simple argument for adverbialism, then: we can represent x even if there is no x; we cannot bear a relation to x if there is no x; therefore, we can represent x even if we do not bear a relation to x. To avoid adverbialism, one would have to deny one of the two premises, but both the claim that we cannot think of dragons and the claim that relations do not require relata are, initially at any rate, much more unpalatable than adverbialism.

The other argument for adverbialism suggested by this cluster of considerations is simply that adverbialism saves us from the can of worms involved in ontological commitment to merely intentional objects. It does so by making the following dual claim (Kriegel 2007): (a) non-derivative dragon-representations are dragon-representations not in virtue of bearing the representation relation to dragons, but in virtue of representing dragon-wise (that is, in virtue of having a dragonesque inferential role or phenomenal character); (b) derivative dragon-representations are dragon representations not in virtue of bearing the representation relation to dragons, but in virtue of bearing the derivation relation (whatever that turns out to be) to non-derivative dragon-representations. In both cases, then, dragon-representations are not such in virtue of bearing a relation to dragons. Thus adverbialism clears us of ontological commitment to dragons (abstract, mental, or non-existent).

In the same vein, only adverbialism can make sense of failure of existential generalization. We all know that from the fact that x represents y it does not follow that there is a y that x represents. It is not often appreciated, however, that from the fact that x bears a relation to y it should very much follow that there is a y that x bears a relation to. This applies to the representation relation as to any other: from the fact that x bears the representation relation to y it follows that there is a y that x bears the representation relation to. However, from the fact that x represents y-wise it does not follow that there is a y. Thus failure of existential generalization is made intelligible by adverbialism.

5 In praise of adverbialism

It may be worth pointing out at this stage that the other mark of intentionality, failure of substitution of co-referential terms, may also offer support for an adverbialist view, although the argument that it does may be more complicated. The idea, in bare outlines, would be this. We all know that from the facts that x represents y and y = z it does not follow that x represents z. But here too, from the facts that x bears relation R to y and that y = z, it does follow that x bears R to z. This applies to the representation relation: from the facts that x bears the representation relation to y and that y = z it follows that x bears the representation relation to z. However, from the facts that x represents y-wise and y = z it does not follow that x represents z-wise. Thus it is possible to think Phosphorus-wise without thinking Hesperus-wise (namely, if one’s thought’s inferential role or phenomenal character is Phosphorescent rather than Hesperescent), even though it is impossible to bear a relation to Phosphorus without bearing it to Hesperus.

Likewise, there should be facts of the matter (to do with inferential role or phenomenal character) that distinguish thinking rabbit-wise from thinking undetached-rabbit-parts-wise, even if it is impossible to bear a relation to rabbits without bearing it to undetached rabbit parts, the two being necessarily coextensive. The representation relation may well be inscrutable in the relevant sense, but the representation modification (i.e., representation as understood by the adverbialist) is very likely “scrutable.”²⁰

Further: it is possible to imagine a Greek without imagining either a blue-eyed Greek or a brown-eyed Greek, even though it is impossible to bear a relation to a Greek without bearing it either to a blue-eyed Greek or to a brown-eyed Greek. What is possible is to imagine Greek-wise without imagining either blue-eyed-Greek-wise or brown-eyed-Greek-wise.

Finally, consider the problem raised by so-called self-locating beliefs or thoughts (Perry 1979).²¹ When John Perry is (unbeknownst to him) making a mess in a store by pushing around a cart with a torn sugar bag, there is a big difference between him thinking that the person with the torn bag is making a mess and him thinking that he himself is making a mess.²² Yet the proposition to which Perry is related on both occasions is one and the same (at least as propositions are traditionally construed). Perry’s own solution to the problem is to claim that beliefs are not individuated only in terms of the propositions they are related to, but also involve constitutively ways of believing. This sounds very much like the introduction of an adverbial element into the account of belief: believing I-am-making-a-mess-wise is different from believing the-person-with-the-torn-bag-is-making-a-mess-wise.²³

Such considerations—to do with substitution failure, inscrutability, non-specificity, self-location, etc.—offer further support for adverbialism, over and above the support it receives from its avoidance of the can of worms of merely intentional objects. Thus it strikes me that adverbialism should be taken seriously in discussions of intentionality. My own preference is for phenomenological adverbialism, but this will not concern us here. In the remainder of this paper, I want to consider a number of objections to adverbialism and to its role in relieving ontological commitment to merely intentional objects.

6 But isn’t adverbialism dead?

It may seem odd that I should advocate an adverbial theory of intentionality, given that few philosophers today seem to take seriously the original adverbial theory of perception (Ducasse 1942; Chisholm 1957). If the theory is so obviously implausible for perception, how can we seriously apply it to all intentionality?

I think there are two main reasons adverbialism about perception is “dead” (sociologically speaking), one technical and one non-technical. The non-technical reason is that, at bottom, the adverbial theory of perception was felt to miss the fact that perception is inherently intentional. Clearly, once we embrace adverbialism as an account of intentionality, this reason for rejecting adverbialism about perception evaporates. For the adoption of an adverbial account of perception is seen to be consistent with construing perception as inherently intentional.

The technical reason for rejecting perceptual adverbialism has to do with a particular objection, due to Jackson (1977), which does threaten to carry over to intentional adverbialism. In its adapted form, the objection would be that while “I am thinking about green dragons and purple butterflies” is not equivalent to “I am thinking about purple dragons and green butterflies,” both would have to be paraphrased by the adverbialist into the same sentence, “I am thinking green-wise and purple-wise and dragon-wise and butterfly-wise.”

The natural move for the adverbialist to make is to propose the following distinguishable renderings: “I am thinking green-dragon-wise and purple-butterfly-wise” for the former and “I am thinking purple-dragon-wise and green-butterfly-wise” for the latter. However, these newer paraphrases lose a level of compositionality present in the original sentences. This may be problematic in a number of ways, but the main one is that the compositionality of the original sentences enabled certain inferences which may be disabled in the paraphrase.²⁴ In particular, the claim is that while from “I am thinking about a green dragon” one can infer “I am thinking about a dragon,” from “I am thinking green-dragon-wise” one cannot infer “I am thinking dragon-wise,” because “dragon-wise” is only a morphological component, not a syntactic component, of “green-dragon-wise” (“green-dragon-wise” being syntactically simple). If this is right, then the adverbial paraphrases, despite purporting to be equivalent to the original sentences, would actually fail to preserve the inferential connections of the latter to surrounding sentences in the logical neighborhood.

It seems to me, however, that while the lack of compositionality blocks one way of making the above inference, other ways remain unblocked. One way to reason from “I am thinking about green dragons” to “I am thinking about dragons” might be through something like conjunction elimination: I am thinking about green dragons; therefore, I am thinking about green and I am thinking about dragons; therefore, I am thinking about dragons. This mode of reasoning is unavailable on the adverbialist story, since one cannot reason as follows: I am thinking green-dragon-wise; therefore, I am thinking green-wise and dragon-wise; therefore, I am thinking dragon-wise. Note, however, that the same fallacy would attach to the following reasoning: I am kicking a strawberry; therefore, I am kicking a straw and I am kicking a berry; therefore, I am kicking a berry. And yet the inference from “I am kicking a strawberry” to “I am kicking a berry” seems acceptable. So the question is what makes it acceptable, and whether the adverbialist cannot co-opt it. Now, it seems that what makes the last inference acceptable is some tacit principle to the effect that the property of being a berry is a determinable of which the property of being a strawberry is a determinate. So the full reasoning is something like this: I am kicking a strawberry; whatever is a strawberry is a berry; therefore, I am kicking a berry. This form of reasoning is valid, and I contend that it is available to the adverbialist as well, because another principle our reasoner could appeal to is that the property of thinking dragon-wise is a determinable of which the property of thinking green-dragon-wise is a determinate. So we can reason as follows: I am thinking green-dragon-wise; whatever is a green-dragon-wise thinking is a dragon-wise thinking; therefore, I am thinking dragon-wise.²⁵

There may be, then, a reasonable response to Jackson’s objection, reasonable enough, certainly, that adverbialism could not be pronounced dead on its strength alone. And since the other major reason for doing so—the thought that perception is inherently intentional—has no bite in the present context, I conclude that adverbialism about intentionality is alive and well.

7 Shouldn’t intentionality connect us to the world?

I have assumed that adverbialism about intentionality is a coherent and intelligible position. But a critic might insist that the whole point of intentionality is that it puts us in cognitive contact with reality. It connects us to the world. The worry is that while the relational property of representing something clearly accomplishes this, the adverbial property of representing somehow does not. If so, intentionality cannot be identified with an adverbial property.

My response has several aspects. First, I want to agree that intentionality connects us to the world, but suggest that it may do so contingently rather than constitutively. As noted in Sect. 3, adverbialism accepts the existence of a representation relation, and allows that many (hopefully most) of our intentional states bear it to worldly entities, denying only that the non-veridical intentional states do as well. Given that the representation relation does manage to connect us to the world, it follows that many of our intentional states connect us to the world. It is just that what makes them the intentional states they are and what makes them connect us to the world are two different properties: they are intentional states because they represent somehow, but connect us to the world because they represent something.

The point of this response is that there is a confusion lurking in the way the “cognitive contact” requirement is often wielded. The requirement need not be satisfied by intentionality as such, only by successful intentionality. A true thought connects us to the world; but it does so in virtue of being true, not in virtue of being a thought. An accurate auditory experience as of trumpets connects us to trumpets, not in virtue of being an experience however, but in virtue of being accurate. Admittedly, there must be a bridge between representation and reality. But representation is the bridged not the bridging. The bridging is provided by the notions of truth, accuracy, veridicality, etc.

What intentionality itself must be required to do is make possible our cognitive contact with the world. I realize that it is a little opaque what “making possible” means, but then again “cognitive contact” itself is not the most transparent of expressions. One way to think of the present point is that intentionality provides truth conditions or accuracy conditions, which then may or may not be satisfied. When the conditions are satisfied, cognitive contact with the world has been established. The role of intentionality is to make such contact possible by laying the conditions whose satisfaction would constitute the establishment of contact.

If this is right, then in order to identify intentionality with the relevant adverbial property, all that must be the shown is that the adverbial property sets conditions whose satisfaction would establish contact with the world. This is not a trivial task, but it is one regarding which the literature already contains some insightful suggestions. For my part, I am particularly fond of Siewert’s (1998) suggestion that a subject is assessable for accuracy purely in virtue of the phenomenal character of her mental states (and without need of any interpretation). This observation is quite compelling and could help phenomenological adverbialism meet its burden.

A related issue is whether adverbialism can accommodate the wide aspect of intentionality. According to externalist views of intentionality, sometimes part of what makes a token act fall under a certain intentional type is facts about the act’s intentional object (Putnam 1975). If so, there is no way to type intentional acts in complete independence of intentional objects, contrary to adverbialism. In essence, my response is to reject externalism about intentionality, therefore deny that intentionality has a wide aspect. Obviously, arguing for this seriously is not a matter for the back of a paper, and must be left for another occasion. I do concede the following conditional: if intentionality has a wide aspect, adverbialism cannot be true of it.

I conclude that when the connection-to-the-world requirement is properly understood, there is reason to expect that adverbialism can meet it.

8 Isn’t adverbialism too revisionary?

Paraphrases come in two varieties. Some paraphrases are conservative: they purport to reveal the “deep grammar” of our everyday manners of speaking. Others are revisionary: they do not pretend to preserve or reveal features of our language but to recommend new ways of speaking, ways purged of the systematic errors our current manners of speaking are. The challenge for a conservative paraphrase is to show that it indeed captures the deep grammar of our everyday manners of speaking, presumably by replicating much of the inferential role of the home sentences. The challenge for a revisionary paraphrase is completely different: to show that the benefit in the new, revolutionary way of speaking justifies the cost of revision. It might be claimed that the adverbial paraphrase of intentional ascriptions recommended here is too radical a departure from everyday intentional talk to qualify as conservative, but at the same time the benefits of adopting it are not significant enough to justify such a departure.

My response to this objection is twofold. First, if the adverbial paraphrase were revisionary, it would still be quite justified. The alternative, after all, is to plunge into the can of worms of merely intentional objects. If our everyday intentional talk is relational rather than adverbial, then our mundane conceptual scheme is committed to the idea that we routinely think of and perceive objects that either do not exist, or are abstract, or are mental. Any of these commitments, it seems to me, is so embarrassing as to invite radicalism. It is quite probable that even proponents of merely intentional objects accept the following conditional: if we can avoid commitment to merely intentional objects, we should. It is just that they deny the antecedent.

The second plank of my response is that the adverbial paraphrase may actually qualify as conservative. Consider the home sentence:

• (H) Jimmy loves Shrek.

And Consider now the following two paraphrases, a relational one and an adverbial one:

• (RP) Jimmy is love-related to Shrek.

• (AP) Jimmy loves Shrek-wise.

The question is: which of the two paraphrases is conservative and which revisionary?²⁶ The objector claims that RP is the conservative paraphrase of H, whereas AP is the revisionary one. I am not so sure. It depends on which of the two preserves the inferential role of H better.

It is important to note that the inferential roles of RP and AP are different. Most importantly, as noted in Sect. 4, RP has existential entailments that AP lacks: the former does, whereas the latter does not, entail that there is this entity Shrek. Now, it is widely accepted that in everyday life H is not taken to have this entailment. So this is a case where AP succeeds in replicating the inferential role of H while RP fails to. Similar remarks can be produced on the basis of the observations made in Sect. 5.

Our objector might retort that there is also a type of entailment that H has and that RP replicates while AP does not. Although it is commonly agreed that H does not entail that there is something that Jimmy loves, H does seem to entail that Jimmy loves something. This is an entailment that RP can recover, since it clearly entails that Jimmy is love-related to something. By contrast, AP seems to fail this test, since it does not entail that Jimmy loves something.²⁷ However, AP does entail that Jimmy loves somehow (in the sense that Jimmy loves something-wise), and this may be taken to recover the original entailment, since according to the adverbialist “Jimmy loves something” requires paraphrase just as much as H anyway. Moreover, it is not altogether unproblematic to hold that H entails that Jimmy loves something but does not entail that there is something Jimmy loves, given that “Jimmy loves something” appears to entail “there is something that Jimmy loves.”

I conclude that there is good reason to suspect that adverbialist paraphrases are in fact conservative rather than revisionary. In addition, I contend that the can of worms attending the assay of merely intentional objects is such that a revisionary paraphrase which would singularly manage to avoid them would be well-justified.

9 What about avoiding merely intentional objects without adverbialism?

I have considered a number of objections to adverbialism being true. I now want to consider objections to adverbialism being required for avoiding commitment to merely intentional objects. I will consider two such objections.

First, there is the view that intentionality is relational, but involves a relation to a content rather than an object. The view is that while some intentional acts have an object and some do not (so that a relation to an object is indeed a contingent matter), all intentional acts necessarily bear a relation to a different kind of entity, called “content.” This entity is most naturally construed as abstract—as a set or collection of properties, as Fregean “concepts” or “thoughts,” as sets of possible worlds, or something in the vicinity—though in principle one could also construe it as a mental concretum. This view may be thought to constitute a non-adverbial account of intentionality that manages to relieve ontological commitment to merely intentional objects.

In response, I want to present a destructive dilemma against this objection: either (a) the “content” of an intentional act is that which the act is about or (b) it is not; if (a), the view does not relieve ontological commitment to merely intentional objects; but if (b), the view is a version of adverbialism rather than an alternative to it.

In many contemporary discussions (especially within the framework of “naturalizing intentionality”), the intentional content of a thought is treated as that which one thinks of/about. We are often told, for example, that representation is a relation between a “vehicle” and a “content,” where the vehicle is what-does-the-representing and the content is what-is-being-represented. On this construal, intentional content is really just intentional object without the confusing effect of the term “object.” Thus given that on the act-content view every thought has a content, if content is that-which-is-being-thought-about (or what-is-being-represented), then there is something that a dragon thought is about. We end up with the same can of worms we have been trying to avoid—no progress has been made.

The only version of the act-content view that may constitute progress, then, construes content as something other than what-is-being-thought-about. But then the relation of an act/state to its content is not the aboutness relation. The relationship between a representational vehicle and a representational content is not, it turns out, the representation relation. So the version of the act-content view under consideration simply denies that the aboutness or representation relation is constitutive of intentionality. This is, in effect, a version of adverbialism. According to inferentialist adverbialism, to think dragon-wise is to have a thought with a dragonesque inferential role, and according to phenomenological adverbialism, it is to have a thought with a dragonesque phenomenal character. What we have on our hands here is a third version of adverbialism, one according to which to think dragon-wise is to have a thought related to a dragonesque content, that is (probably), to bear a relation to the set of all properties associated with dragons. We may call this view content adverbialism.²⁸

I conclude that under no interpretation does the act-content theory constitute an alternative to adverbialism that relieves commitment to merely intentional objects. Under one interpretation, it does not relieve such commitment at all; under another, it does but is not an alternative to adverbialism.

A second objection of the same form is that the act-object theory of intentionality can avoid merely intentional objects by denying the possibility of dragon thoughts, fountain-of-youth desires, butterfly hallucinations, etc. If such intentional states are impossible, then all intentional states may have unproblematic objects.

I want to concede that the act-object theory can also avoid commitment to merely intentional objects, in the manner suggested, but it should be fully realized what the price is. On the position under consideration, someone who undertakes to think that Sherlock Holmes is a unicorn is bound to fail to have any thought. In attempting to have thoughts we are beholden to the world’s cooperation: if the world cooperates, we will manage to think; if it does not, the wheels of thought will be cranking and chomping furiously but to no consequence.²⁹ There is no hallucination and no imagination, and no priest has ever managed to have a Godly thought. Things get stranger yet when we move to apparent intentional states with a world-to-mind rather than mind-to-world direction of fit. Thus the position under consideration has the happy consequence that everybody’s desires are always satisfied. We do not have desires that are frustrated, only failed attempts to have desires. To my mind, all this is quite absurd.

In any case, the two objections considered in this section do not threaten the idea that there are no merely intentional objects. Eliminativism about such putative entities is granted by the objector. The only question is whether the kind of phenomenological adverbialism I favor is the only ticket to eliminativism. The answer is that there may be other tickets, in the form of content adverbialism and an act-object theory that rejects thoughts of non-existents, but these tickets are disproportionately expensive.

10 Conclusion

Intentional objects form the highest and most inclusive category of entities we might end up ontologically committing to. Ontological commitment to intentional objects, including merely intentional objects, would entail commitment to mathematical entities, moral properties, events, possibilia, and any other putative entity that is thinkable. It is therefore the first order of business for any philosopher with a taste for desert landscapes or a general fear of commitment to undermine the case for (merely) intentional objects. This paper proposed a way to do so, one that involves an adverbial theory of intentionality. I have attempted to elucidate adverbialism somewhat, motivate it somewhat, and defend it against a number of foundational worries. Much work remains to be done in pursuit of a more articulated adverbialist program. My hope is to have made the case that this may be worth the effort.